[development] Code freeze?
drupal-devel at webchick.net
Thu Jun 26 13:56:52 UTC 2008
Dries Buytaert wrote:
> I do think we're doing great work on testing. I've been committing
> testing related patches on a (almost) daily basis. That's pretty sweet.
Like catch and others, I've been dumping basically all the time I have
available to work on D7 on testing stuff, in the hopes that it will help
extend the code freeze out to when I will have much MORE time to dump
into D7 in the fall. ;)
> That said, we aren't able to measure our test coverage yet. In other
> words, it's really hard to tell how well we actually do. Any updates on
I think that our current test coverage by conventional tools is going to
be close to 0%. It's also clear that the community has not in fact
"embraced testing." Instead, a hardcore group of around 15-20 people
(the "testing brigade" ;)) have, and are driving this effort forward on
behalf of the rest of the development team.
I also think that both of these situations are okay right now. Because
what the "testing brigade" has been focusing on is:
1. Improvement of testing tools. For example, that Batch API patch that
was just committed I think is the *key* turning point that will make
testing something that can be done by normal humans rather than just the
2. Fixing of existing tests. Back when I wrote
http://webchick.net/itch-of-the-week/fix-testing-crisis, we were in
pretty sorry shape in this regard. I remember growing the critical bug
queue by at least 20 in one night documenting all the tests that didn't
pass. Now, we're down to 3 failing tests as of this morning.
3. Improving coverage of tests that run through the end-user experience
via the browser. This helps save our reviewers from getting carpal
tunnel clicking on forms to ensure that the basic system is running
properly while they're testing a patch.
4. Developing testing guidelines, best practices. We're not totally
there yet, but a fairly large amount of time has been spent on things
like documentation, clean-up of tests so that they all follow basically
the same conventions, etc. This type of work is important to getting new
developers on the testing train.
However, all of this "clean-up" work has come at the expense of
dramatically increasing our testing coverage. But I actually think that
it would've been way too premature to dump a bunch of time into
increasing our test coverage while the above 4 points were outstanding.
My goal is for the "Awesome Testing Party" at Drupalcon Szeged (if it
gets accepted) to help a lot in this regard.
More information about the development